TCSinger: Zero-Shot Singing Voice Synthesis with Style Transfer and Multi-Level Style Control

TL;DR

TCSinger is a pioneering zero-shot singing voice synthesis model that enables style transfer and multi-level style control across languages and singing methods, producing high-quality, stylistically nuanced singing voices from audio and text prompts.

Contribution

It introduces a novel multi-module framework for zero-shot style transfer in singing voice synthesis, including style encoding, style and duration prediction, and style adaptive decoding.

Findings

Outperforms baseline models in quality and style similarity

Effective cross-lingual and speech-to-singing style transfer

Enables detailed multi-level style control

Abstract

Zero-shot singing voice synthesis (SVS) with style transfer and style control aims to generate high-quality singing voices with unseen timbres and styles (including singing method, emotion, rhythm, technique, and pronunciation) from audio and text prompts. However, the multifaceted nature of singing styles poses a significant challenge for effective modeling, transfer, and control. Furthermore, current SVS models often fail to generate singing voices rich in stylistic nuances for unseen singers. To address these challenges, we introduce TCSinger, the first zero-shot SVS model for style transfer across cross-lingual speech and singing styles, along with multi-level style control. Specifically, TCSinger proposes three primary modules: 1) the clustering style encoder employs a clustering vector quantization model to stably condense style information into a compact latent space; 2) the Style and Duration Language Model (S\&D-LM) concurrently predicts style information and phoneme duration, which benefits both; 3) the style adaptive decoder uses a novel mel-style adaptive normalization method to generate singing voices with enhanced details. Experimental results show that TCSinger outperforms all baseline models in synthesis quality, singer similarity, and style controllability across various tasks, including zero-shot style transfer, multi-level style control, cross-lingual style transfer, and speech-to-singing style transfer. Singing voice samples can be accessed at https://aaronz345.github.io/TCSingerDemo/.